List your company

Scientific Reasoning on ResearchQA (test)

73.1AccuracyRCSD
62.80465.47768.1570.823Jun 17, 2026

Evaluation results

MethodLinks
RCSDBackbone=Qwen3-8B, Training Steps=1002026.0673.1
OPSDBackbone=Qwen3-8B, Training Steps=1002026.0672.8
Rubric-GRPOBackbone=Qwen3-8B2026.0672.1
GRPOBackbone=Qwen3-8B, Training Steps=5002026.0671.5
Qwen3-8BBackbone=Qwen3-8B2026.0664.9
SFTBackbone=Qwen3-8B2026.0663.2